Tutorial 8.pdf - Vula

Quantitative Methods for EconomicsTutorial 8Katherine Eyal

(b) Why might nox [or more precisely, log (nox)] and rooms be negatively correlated?If this is the case, does the simple regression of log (price) on log (nox)produce an upward or a downward biased estimator of β 1 ?(c) An empirical analysis yields the following results:log ̂(price) = 11.71 − 1.043 log (nox) , n = 506, R 2 = 0.264log ̂(price) = 9.23 − 0.718 log (nox) + 0.306 rooms, n = 506, R 2 = 0.514Is the relationship between the simple and multiple regression estimates of theelasticity of price with respect to nox what you would have predicted, givenyour answer in part (b)? Does this mean that −0.718 is definitely closer to thetrue elasticity than −1.043?Part B: Computer Exercises1. How do growing weather and a wine’s age influence a Bordeaux wine’s price? Thedata set WINEWEATHER1.DTA contains average 1983 prices for Bordeaux winesfor the vintages from 1952 to 1980, along with data on weather conditions when eachvintage was being grown. These data were part of an analysis of Bordeaux wine as aninvestment by economists Orley Ashenfelter, David Ashmore, and Robert LaLonde.The variables in the file arevint Vintage of the wine (i.e. its year of production)logprice Natural log of the price of Bordeaux wines relative to the price ofthe 1961 vintagedegrees Average temperature in the growing seasonhrain Rainfall in the harvest seasonwrain Winter rainfall prior to harvest seasontime sv Time from 1983 back to the wine’s vintage year(a) Use multiple regression to explore how growing-season temperatures, harvestseasonrainfall, off-season rainfall, and the age of a wine influence the naturallog of a vintage’s price.(b) Are the signs on the variables what you expect? Briefly explain.(c) How much of the variation in vintages’ prices in this sample is accounted for bythese explanatory variables?(d) Regress the log of price on the age of the wine and an intercept term. How doyou interpret the coefficients on age in this regression and in the regression in(a)?(e) Why does the variable logprice have the value zero for the 1961 vintage?2

2. Use the data in CHARITY.DTA to answer the following questions:(a) Estimate the equationgift = β 0 + β 1 mailsyear + β 2 giftlast + β 3 propresp + uby OLS. How does the R-squared compare with that from the simple regressionthat omits giftlast and propresp (you estimated this simple regression inTutorial 6)?(b) Interpret the coefficient on mailsyear. Is it bigger or smaller than the correspondingsimple regression coefficient?(c) Interpret the coefficient on propresp. Be careful to notice the units of measurementof propresp.(d) Now add the variable avggift to the equation. What happens to the estimatedeffect of mailsyear?(e) In the equation from part (d), what has happened to the coefficient on giftlast?What do you think is happening?3. Use the data in WAGE1.DTA to confirm the partialling out interpretation of theOLS estimates by explicitly doing the partialling out for Example 3.2 on page 76 ofWooldridge. This first requires regression educ on exper and tenure and saving theresiduals, ̂r 1 . Then, regress log(wage) on ̂r 1 . Compare the coefficient on ̂r 1 with thecoefficient on educ in the regression of log(wage) on educ, exper, and tenure.3

TUTORIAL 8 SOLUTIONS27 September 2010ECO3021SPart A: Problems1. The following model is a simplified version of the multiple regression model used byBiddle and Hamermesh (1990) 1 to study the trade-off between time spent sleepingand working and to look at other factors affecting sleep:sleep = β 0 + β 1 totwrk + β 2 educ + β 3 age + uwhere sleep and totwrk (total work) are measured in minutes per week and educ(years of education) and age are measured in years.(a) If adults trade off sleep for work, what is the sign of β 1 ?(b) What signs do you think β 2 and β 3 will have?(c) Using the data in Biddle and Hamermesh (1990), the estimated equation isŝleep = 3, 638.25 − 0.148 totwrk − 11.13 educ + 2.20 agen = 706, R 2 = 0.113If someone works five more hours per week, by how many minutes is sleeppredicted to fall? Is this a large trade-off?(d) Discuss the sign and magnitude of the estimated coefficient on educ.(e) Would you say totwrk, educ and age explain much of the variation in sleep?What other factors might affect the time spent sleeping? Are these likely to becorrelated with totwrk?SOLUTION:(a) If adults trade off sleep for work, more work implies less sleep (other thingsequal), so β 1 < 0.1 J.E. Biddle and D.S. Hamermesh (1990), “Sleep and the Allocation of Time,” Journal of PoliticalEconomy 98, 922-943.1

(b) The signs of β 2 and β 3 are not obvious, at least to me. One could argue that moreeducated people like to get more out of life, and so, other things equal, they sleepless (β 2 < 0). The relationship between sleeping and age is more complicatedthan this model suggests, and economists are not in the best position to judgesuch things.(c) Since totwrk is in minutes, we must convert five hours into minutes: ∆totwrk =5(60) = 300. Then sleep is predicted to fall by 0.148(300) = 44.4 minutes. Fora week, 45 minutes less sleep is not an overwhelming change.(d) More education implies less predicted time sleeping, but the effect is quite small.If we assume the difference between college and high school is four years, thecollege graduate sleeps about 45 minutes less per week, other things equal.(e) Not surprisingly, the three explanatory variables explain only about 11.3% ofthe variation in sleep. One important factor in the error term is general health.Another is marital status, and whether the person has children. Health (howeverwe measure that), marital status, and number and ages of children wouldgenerally be correlated with totwrk. (For example, less healthy people wouldtend to work less.)2. The following equation describes the median housing price in a community in termsof amount of pollution (nox for nitrous oxide) and the average number of rooms inhouses in the community (rooms):log (price) = β 0 + β 1 log (nox) + β 2 rooms + u(a) What are the possible signs of β 1 and β 2 ? What is the interpretation of β 1 ?Explain.(b) Why might nox [or more precisely, log (nox)] and rooms be negatively correlated?If this is the case, does the simple regression of log (price) on log (nox)produce an upward or a downward biased estimator of β 1 ?(c) An empirical analysis yields the following results:log ̂(price) = 11.71 − 1.043 log (nox) , n = 506, R 2 = 0.264log ̂(price) = 9.23 − 0.718 log (nox) + 0.306 rooms, n = 506, R 2 = 0.514Is the relationship between the simple and multiple regression estimates of theelasticity of price with respect to nox what you would have predicted, givenyour answer in part (b)? Does this mean that −0.718 is definitely closer to thetrue elasticity than −1.043?2

SOLUTION:(a) β 1 < 0 because more pollution can be expected to lower housing values; notethat β 1 is the elasticity of price with respect to nox. β 2 is probably positivebecause rooms roughly measures the size of a house. (However, it does notallow us to distinguish homes where each room is large from homes where eachroom is small.)(b) If we assume that rooms increases with quality of the home, then log(nox)and rooms are negatively correlated when poorer neighborhoods have morepollution, something that is often true. We can use Table 3.2 on page 91 ofWooldridge to determine the direction of the bias. If β 2 > 0 and Corr(x 1 , x 2 )

(b) Are the signs on the variables what you expect? Briefly explain.(c) How much of the variation in vintages’ prices in this sample is accounted for bythese explanatory variables?(d) Regress the log of price on the age of the wine and an intercept term. How doyou interpret the coefficients on age in this regression and in the regression in(a)?(e) Why does the variable logprice have the value zero for the 1961 vintage?SOLUTION:(a) Command: reg logprice wrain degrees hrain time svModel 8.66443586 4 2.16610897 Prob > F 0Residual 1.80582883 22 .082083129 R-squared 0.8275Adj R-squared 0.7962Total 10.4702647 26 .402702488 Root MSE 0.2865logprice Coef. Std. Err t P>t 95% Conf. Intervalwrain 0.0011668 .000482 (2.42) 2.40% 0.0001671 0.0021665degrees 0.6163926 .0951755 (6.48) 0.00% 0.4190107 0.8137745hrain −0.0038606 .0008075 (−4.78) 0.00% −0.0055353 −0.0021858time sv 0.0238474 .0071667 (3.33) 0.30% 0.0089846 0.0387103cons −12.14534 1.688103 (−7.19) 0.00% −15.64625 −8.644426wrain: If winter rain increases by 100ml then the average relative price of Bordeauxwines increase by 11.66%degrees: If the average temperature increases by 1 deg centigrade then averagerelative price of Bordeaux wines increases by 62%hrain: If harvest rainfall increases by 100ml then average relative price of Bordeauxwines decreases by 38.6%time sv : This variable gives the age of the wine. If the age of the wine increasesby 1 year then average relative price of Bordeaux wines increases by 2.4%Note that all the variables are significant at the 1% level except wrain which issignificant at the 5% level.(b) The signs are as expected one would expect winter rain fall, warm weather andthe vintage to increase the value of the average relative price. Summer rainfall is likely to have a negative effect on crop yields and thus the relative price(given that it is uncharacteristic of most wine growing regions to have summerrain fall). The size of the coefficient on degrees is larger than the others, butone must consider that degrees is an average which probably has a low degreeof variation form season to season.4

(c) Look at the R 2 : The explanatory variables explain roughly 83% of the variationin vintages relative prices.(d) Command: reg logprice time svlogprice Coef. Std. Err. t P>ttime sv 0.0354296 0.0136625 2.59 0.016cons -2.025199 0.2472287 -8.19 0Note that one can use time sv or vint to indicate the age of the wine (thesetwo variables are perfectly collinear). Using vint will result in the signs beingreversed (test this). Here the premium for an additional year of the vintageis now higher at 3.5% per year. The difference is now we do not control forall other factors affecting the average relative price as we did in the multipleregression, and they are thus in the error.Note this leads to an upward bias, which is quite significant yet it is unclear whytime sv would be correlated with any of the other above explanatory variablesshould they be in the error but the “other” explanatory variables are certainlycorrelated with y. The most important difference is in the multiple regressionwe have the effect of an additional vintage year on average relative price whilecontrolling for winter and harvest rain fall as well as the average temperature.(e) The price of the 1961 vintage relative to the price of the 1961 vintage is 1, andthe log of 1 is zero.2. Use the data in CHARITY.DTA to answer the following questions:(a) Estimate the equationgift = β 0 + β 1 mailsyear + β 2 giftlast + β 3 propresp + uby OLS. How does the R-squared compare with that from the simple regressionthat omits giftlast and propresp (you estimated this simple regression inTutorial 6)?(b) Interpret the coefficient on mailsyear. Is it bigger or smaller than the correspondingsimple regression coefficient?(c) Interpret the coefficient on propresp. Be careful to notice the units of measurementof propresp.(d) Now add the variable avggift to the equation. What happens to the estimatedeffect of mailsyear?(e) In the equation from part (d), what has happened to the coefficient on giftlast?What do you think is happening?5

SOLUTION:(a) The estimated equation isĝift = −4.55 + 2.17 mailsyear + .0059 giftlast + 15.36 proprespn = 4, 268, R 2 = .0834The R-squared is now about .083, compared with about .014 for the simpleregression case. Therefore, the variables giftlast and propresp help to explainsignificantly more variation in gifts in the sample (although still just over eightpercent).(b) Holding giftlast and propresp fixed, one more mailing per year is estimatedto increase gifts by 2.17 guilders. The simple regression estimate is 2.65, sothe multiple regression estimate is somewhat smaller. Remember, the simpleregression estimate holds no other factors fixed.(c) Because propresp is a proportion, it makes little sense to increase it by one.Such an increase can happen only if propresp goes from zero to one. Instead,consider a .10 increase in propresp, which means a 10 percentage point increase.Then, gift is estimated to be 15.36(.1) ≈ 1.54 guilders higher.(d) The estimated equation isĝift = −7.33 + 1.20 mailsyear − .261 giftlast + 16.20 propresp + .527 avggiftn = 4, 268, R 2 = .2005After controlling for the average past gift level, the effect of mailings becomeseven smaller: 1.20 guilders, or less than half the effect estimated by simpleregression.(e) After controlling for the average of past gifts – which we can view as measuringthe “typical” generosity of the person and is positively related to the currentgift level – we find that the current gift amount is negatively related to the mostrecent gift. A negative relationship makes some sense, as people might follow alarge donation with a smaller one.3. Use the data in WAGE1.DTA to confirm the partialling out interpretation of theOLS estimates by explicitly doing the partialling out for Example 3.2 on page 76 ofWooldridge. This first requires regression educ on exper and tenure and saving theresiduals, ̂r 1 . Then, regress log(wage) on ̂r 1 . Compare the coefficient on ̂r 1 with thecoefficient on educ in the regression of log(wage) on educ, exper, and tenure.6

SOLUTION:The regression of educ on exper and tenure yieldseduc = 13.57 − .074 exper + .048 tenure + ̂r 1n = 526, R 2 = .101.Now, when we regress log(wage) on ̂r 1 we obtainlog (wage) = 1.62 + .092 ̂r 1n = 526, R 2 = .207As expected, the coefficient on in the second regression is identical to the coefficienton educ in equation (3.19) on page 76 of Wooldridge. Notice that the R-squared fromthe above regression is less than that in (3.19). In effect, the regression of log(wage)on ̂r 1 explains log(wage) using only the part of educ that is uncorrelated with experand tenure; separate effects of exper and tenure are not included.7